JMB MS 1679 [15/1/97]

J. Mol. Biol. (1997) 265, 541±552

Mge1 Functions as a Nucleotide Release Factor for Ssc1, a Mitochondrial of

Bingjie Miao, Julie E. Davis and Elizabeth A. Craig*

Department of Biomolecular Mge1, a GrpE-related protein in the mitochondrial matrix of the budding Chemistry, 1300 University yeast Saccharomyces cerevisiae, is required for translocation of precursor Avenue, University of proteins into mitochondria. The effect of Mge1 on nucleotide release from Wisconsin, Madison, WI 53706 Ssc1, an Hsp70 of the mitochondrial matrix, was analyzed. The release of USA both ATP and ADP from Ssc1 was stimulated in the presence of Mge1, therefore we conclude that Mge1 functions as a nucleotide release factor for Ssc1. Mge1 bound stably to Ssc1 in vitro; this interaction was resistant to high concentrations of salt but was disrupted by the addition of ATP. ADP was much less effective in releasing Mge1 from Ssc1 whereas ATPgS and AMPPNP could not disrupt the Ssc1/Mge1 complex. Ssc1-3, a temperature sensitive SSC1 mutant protein, did not form a detectable complex with Mge1. Consistent with the lack of a detectable interaction, Mge1 did not stimulate nucleotide release from Ssc1-3. A conserved loop structure on the surface of the ATPase domain of DnaK has been impli- cated in its interaction with GrpE. Since the single amino acid change in Ssc1-3 lies very close to the analogous loop in Ssc1, the role of this loop in the Ssc1:Mge1 interaction was investigated. Deletion of the loop abol- ished the physical and functional interaction of Ssc1 with Mge1, suggesting that the loop in Ssc1 is also important for the Ssc1:Mge1 inter- action. Two mutants with single amino acid changes within the loop did not eliminate the stable binding of Mge1, yet the binding of Mge1 did not stimulate the release of nucleotides from the mutant SSC1 proteins. We propose that the loop region of Ssc1 is important for the physical interaction between Mge1 and Ssc1, and for generation of a confor- mational change necessary for Mge1-induced nucleotide release. # 1997 Academic Press Limited *Corresponding author Keywords: Hsp70; Mge1; Ssc1; nucleotide release;

Introduction terminal ATPase domain, which binds and hydro- lyzes ATP, and a somewhat less conserved pep- The 70-kDa heat shock proteins () have tide binding domain (Chappell et al., 1987; Wang been highly conserved during evolution and are et al., 1993; Freeman et al., 1995). The tertiary struc- present in every organism examined so far. ture of the 44-kDa ATPase domain is similar to Hsp70s are important for a variety of cellular func- that of hexokinase and actin (Flaherty et al., 1990, tions, including , protein transloca- 1991; Bork et al., 1992), whereas the C-terminal tion across biological membranes, and protein peptide binding domain has a unique b-sandwich degradation (for reviews, see Craig et al., 1993; structure followed by an extended structure of a- Hartl, 1996; Morimoto et al., 1994). Hsp70s have helices (Zhu et al., 1996). The interaction between two functional domains, a highly conserved N- the two domains is critical for the function of Hsp70s (Buchberger et al., 1994b, 1995). Hsp70s function as molecular chaperones by binding to Abbreviations used: ATPgS, adenosine 50-O-(3- short stretches of hydrophobic peptide sequences thiotriphosphate); AMPPNP, 50-adenylyl-b,g- imidodiphosphate; 5-FOA, 5-¯uoroorotic acid; GST, thus preventing premature folding or aggregation glutathione S-transferase; IPTG, isopropyl b-D- of partially unfolded proteins (Flynn et al., 1989; thiogalactopyranoside; PEI-cellulose, polyethyleneimine- Blond-Elguindi et al., 1993; Gragerov et al., 1994). cellulose; TLC, thin layer chromatography. Upon ATP binding and/or hydrolysis, bound

0022±2836/97/050541±12 $25.00/0/mb960762 # 1997 Academic Press Limited JMB MS 1679 [15/1/97]

542 Mge1, a Nucleotide Release Factor for Ssc1

peptide is released to allow for its proper folding Gambill et al., 1993). Ssc1 has been found to be as- (Palleros et al., 1993; McCarty et al., 1995; Banecki sociated with precursor proteins during and after & Zylicz, 1996). This cycle of ATP binding and translocation (Ostermann et al., 1990; Scherer et al., hydrolysis coupled to peptide binding and release 1990; Manning-Krieg et al., 1991), consistent with is essential for the function of Hsp70s. its role in the import and maturation of precursor Several lines of evidence indicate that DnaK, an proteins. The binding of Ssc1 to precursor proteins Hsp70 protein of Escherichia coli, functions together in transit across the mitochondrial membranes is with two proteins, DnaJ and GrpE (reviewed by essential for conferring the unidirectionality of the Georgopoulos et al., 1994). Mutations in the dnaK, import process (Ungermann et al., 1994). Mge1 can dnaJ or grpE result in similar phenotypes; be quantitatively co-immunoprecipitated with Ssc1 furthermore, DnaK, DnaJ and GrpE function from isolated mitochondria, and both Ssc1 and together in a variety of in vitro assays, such as in- Mge1 can be co-immunoprecipitated with a pre- itiation of l DNA replication and refolding of de- cursor protein that is trapped at the import site natured proteins. DnaJ and GrpE exert their (Voos et al., 1994), suggesting a functional co- effects, at least in part, by modulating the ATPase operation between Ssc1 and Mge1 in the process activity of DnaK. DnaJ stimulates the hydrolysis of of protein translocation. bound ATP by DnaK, whereas GrpE promotes the Because of the sequence similarity between Mge1 release of nucleotides from DnaK (Liberek et al., and GrpE, and the functional cooperation between 1991a). DnaJ alone stimulates the steady-state Ssc1 and Mge1, it has been hypothesized that ATPase activity of DnaK by two to tenfold, Mge1 functions as a nucleotide release factor for whereas GrpE alone has a minimal effect (Jordan Ssc1. Here we show that Mge1 is in fact a & McMacken, 1995; McCarty et al., 1995). How- nucleotide release factor for Ssc1. We also charac- ever, DnaJ and GrpE together can stimulate the terized the Ssc1:Mge1 interaction and examined steady-state ATPase activity of DnaK by up to the effect of mutations of the loop in Ssc1, which 100-fold. is analogous to the GrpE-interacting loop in DnaK, Unlike the DnaK:DnaJ interaction, GrpE binds on the interaction of Ssc1 with Mge1. Based on tightly to DnaK. The complex between DnaK and these results, a possible mechanism for Mge1- GrpE is stable in the presence of high concen- induced nucleotide release is discussed. trations of salt, but is disrupted upon the addition of ATP (Zylicz et al., 1987). GrpE binds to the 44- Results kDa ATPase domain of DnaK, and a conserved loop structure on the surface of the ATPase Mge1 is a nucleotide release factor for Ssc1 domain has been implicated in the interaction of DnaK with GrpE (Buchberger et al., 1994a). A Since GrpE functions as a nucleotide release factor point mutation in this loop as well as a deletion of for DnaK, we wanted to test whether Mge1 is this loop eliminates the physical and functional able to function as a nucleotide release factor for interaction between DnaK and GrpE. It has been Ssc1. Consistent with the lack of an effect of GrpE proposed that the binding of GrpE induces a con- on the ATPase activity of DnaK (Jordan & formational change in DnaK, thus triggering McMacken, 1995), Mge1 had a minimal effect on nucleotide release (Buchberger et al., 1994a), yet the ATPase activity of Ssc1 (data not shown). To the mechanism of GrpE-induced nucleotide release examine more closely the interaction of Mge1 with remains to be elucidated. Ssc1, an isolated step in the ATPase reaction, the Numerous eukaryotic Hsp70s and several DnaJ- ATP hydrolysis step, was analyzed by single turn- related proteins have been identi®ed in eukaryotic over experiments. Complexes of Ssc1 and 32 cells. Genetic and biochemical studies have estab- [a- P]ATP were formed at 30C; isolation of the lished the functional interaction between them complex by size exclusion chromatography was (reviewed by Cyr et al., 1994). Mge1 (also referred carried out at 4C to limit the hydrolysis of to as Yge1p, GrpEp), a GrpE-related protein, has [a-32P]ATP. The isolated Ssc1/ATP complex was recently been identi®ed in the mitochondrial then incubated at 30C and the hydrolysis of matrix of Saccharomyces cerevisiae (Laloraya et al., bound ATP monitored. As shown in Figure 1A, 1994; Bolliger et al., 1994; Ikeda et al., 1994). Mge1, wild-type Ssc1 hydrolyzed bound ATP with about which shares 34% identity with Escherichia coli 50% of the bound ATP being hydrolyzed within GrpE (Laloraya et al., 1994), is essential for the four minutes of incubation. To investigate the growth of S. cerevisiae. Mge1 is required for nor- stability of the interaction of ATP with Ssc1, the mal import and maturation of nuclear-encoded ability of excess unlabeled ATP or ADP to quench mitochondrial proteins (Laloraya et al., 1994, 1995; the hydrolysis of the bound [a-32P]ATP was deter- Westerman et al., 1995). mined. Assuming that there is no signi®cant coop- Ssc1, an essential Hsp70 in the mitochondrial erativity in nucleotide binding to Ssc1, addition of matrix, is also required for the translocation and unlabeled nucleotide should have no effect on the maturation of precursor proteins, as temperature- hydrolysis of ATP which remains bound to Ssc1. sensitive mutations in SSC1 result in a block of However, [a-32P]ATP which is released prior to import of precursor proteins upon shifting to the hydrolysis must then compete with the vast excess non-permissive temperature (Kang et al., 1990; of unlabeled nucleotide for rebinding before JMB MS 1679 [15/1/97]

Mge1, a Nucleotide Release Factor for Ssc1 543

Figure 1. Effects of excess nucleotides and Mge1 on the single turnover of Ssc1/ATP complex. A, The Ssc1/ATP complex ( 2 mM) was incubated at 30C with the ad- Figure 2. Mge1 releases both ATP and ADP from Ssc1. dition of Mge1 (4 mM), ATP (250 mM) and ADP (250 mM) A, The Ssc1/ATP complex ( 3 mM) was incubated with  as indicated. Aliquots were withdrawn at the indicated or without Mge1 (30 mM) at 30C to achieve 50 to 60% time points and the fraction of ATP converted to ADP conversion of ATP to ADP before loading onto a G-50 determined as described in Materials and Methods. B, column. Aliquots were collected and counted. Percen- Same as in A with the indicated amount of Mge1 tage of total counts in each fraction was plotted. The added. ®rst peak of radioactivity around fraction 6 corresponds to bound nucleotides whereas the second peak of radioactivity around fraction 14 corresponds to free nucleotides. B, Aliquots from peak fractions were mixed hydrolysis can occur. The rate of hydrolysis of the with stop solution immediately after emerging from the bound [a-32P]ATP was not signi®cantly affected column and assayed for relative amount of ATP and ADP. Percentages of free and bound ATP and ADP by the addition of excess unlabeled ATP or ADP were calculated from relative peak areas and ATP/ADP (Figure 1A), indicating that the bound ATP was ratio in the peak fractions. stable in the time frame of the experiment. At longer time points, excess unlabeled ATP or ADP quenched the hydrolysis of bound [a-32P]ATP, in- dicating that bound ATP was slowly released decrease the rate of hydrolysis further. Such a from Ssc1 (data not shown). decrease was observed when the ratio of However, when Mge1 was added to the reaction, Mge1:Ssc1 was increased from 2:1 to 10:1 a signi®cant drop in the rate of hydrolysis of (Figure 1B). [a-32P]ATP was observed. Furthermore, when To test more directly the effect of Mge1 on the excess cold ATP or ADP was added together with release of bound nucleotide from Ssc1, isolated 32 Mge1, the hydrolysis of bound [a- P]ATP was Ssc1/ATP complex was incubated at 30C in the almost completely quenched (Figure 1A). Together absence or presence of Mge1 to achieve 50 to 60% these results indicate that addition of Mge1 causes conversion of [a-32P]ATP to [a-32P]ADP. The reac- release of bound ATP from Ssc1. In the presence tion mixture was again subjected to size exclusion of excess cold ATP or ADP, the released chromatography to monitor the release of bound [a-32P]ATP was rarely rebound thus the hydrolysis ATP/ADP. Although a signi®cant amount of of [a-32P]ATP was almost totally quenched. In the radioactivity remained associated with Ssc1 in the absence of excess cold competitors, the released absence of Mge1, a peak of free nucleotides was [a-32P]ATP could be rebound by Ssc1, but this also observed (Figure 2A). Analysis of the peak release and rebinding would slow down the rate fractions revealed that 81% of total ATP remained of hydrolysis. If this hypothesis is true, one would bound to Ssc1, whereas only 13% of total ADP predict that if the concentration of Mge1 wasn't was still bound to Ssc1. The free nucleotide peak already saturating, addition of more Mge1 would contains the remaining 19% of ATP and 87% of JMB MS 1679 [15/1/97]

544 Mge1, a Nucleotide Release Factor for Ssc1

Figure 3. Effects of adenine nucleotides on the stability of the Ssc1/Mge1 complex. GST-Ssc1 fusion protein was im- mobilized onto glutathione agarose beads and puri®ed Mge1 (T) was mixed with the beads. After incubation at 4C for 30 minutes the unbound fraction was collected (FT). The beads were washed with buffer containing 1 M NaCl (SW), and a sample of the beads was taken (B1). The beads were then eluted with the indicated concentrations of ade- nine nucleotides (E) and a sample of the beads after elution was taken (B2). Immunoblots with anti-Mge1 antibody are shown.

ADP (Figure 2B, lane 1 and lane 3). This distri- periment was performed with sub-stoichiometric bution of nucleotide suggests that ADP is prefer- levels of Mge1, with the Ssc1:Mge1 ratio being ably released from Ssc1, and ATP is relatively 10:1. Control experiments showed that when stably bound to Ssc1 during the time required to excess Mge1 was added, the immobilized GST- collect the bound fractions ( eight minutes at Ssc1 fusion protein was capable of retaining ap-  4C). This stability of bound ATP is consistent proximately an equal molar amount of Mge1 (data with the results shown in Figure 1A, as excess not shown). unlabeled ATP did not quench the hydrolysis To test the stability of the Ssc1:Mge1 interaction, of prebound [a-32P]ATP. However, when Mge1 the beads were then washed extensively with 1 M was added, a signi®cant drop in the amount of NaCl. No Mge1 was detected in the salt wash radioactivity associated with Ssc1 was observed fraction, indicating that the interaction is stable in (Figure 2A), supporting the idea that Mge1 stimu- the presence of high concentrations of salt. The lates the release of nucleotides from Ssc1. Further- susceptibility of the Mge1:Ssc1 interaction to more, in the presence of Mge1, the fraction of total nucleotides was also tested. Incubation in the pre- ATP remaining associated with Ssc1 dropped sence of 1 mM ATP resulted in the release of the from 81 to 47% (Figure 2B, lane 1 versus lane 2), majority of the bound Mge1. The portion of Mge1 whereas bound ADP dropped from 13 to only 3% that remained in the beads fraction was probably (Figure 2B, lane 3 versus lane 4). This result indi- due to incomplete separation by batch elution, cates that Mge1 promotes the release of both ATP since it was signi®cantly diminished by more and ADP from Ssc1, and thus functions as a extensive washes with ATP (data not shown). As nucleotide release factor for Ssc1. little as 10 mM ATP was effective in releasing Mge1 from Ssc1. ADP was much less effective in Mge1 can stably associate with Ssc1 releasing Mge1 from Ssc1, requiring a concen- tration of 10 mM for signi®cant release of Mge1. Since Mge1 could function as a nucleotide release This release by high concentrations of ADP is not factor for Ssc1, we asked whether interaction due to minor ATP contamination since pretreating between the two proteins could be detected the ADP with hexokinase had no effect on the directly. We took advantage of the fact that the release of Mge1 by ADP (data not shown). ATPgS GST-Ssc1 fusion protein, which we constructed for and AMPPNP were not able to effectively release puri®cation of Ssc1 (see Materials and Methods), Mge1 from Ssc1 (Figure 3B). The faint bands that could be immobilized onto glutathione agarose appeared in the 10 mM elution lanes could be beads. After binding of GST-Ssc1 fusion protein to attributed to ATP contamination in the commer- the glutathione af®nity beads, puri®ed Mge1 was cially available ATPgS and AMPPNP (Ahsen et al., mixed with the immobilized Ssc1. As shown in 1995; Horst et al., 1996). The failure of these ATP Figure 3A, essentially all the Mge1 was retained analogs to release Mge1 suggests that either the on the beads. In the absence of the GST-Ssc1 analogs do not bind to Ssc1 because of lower af®- fusion protein, no retention of Mge1 on the beads nities as reported for Hsc70 (Gao et al., 1994), or was observed, indicating that the interaction with the binding of the analogs is different from the the GST-Ssc1 fusion protein was speci®c. This ex- binding of ATP, as reported for DnaK (Liberek JMB MS 1679 [15/1/97]

Mge1, a Nucleotide Release Factor for Ssc1 545

This lack of co-immunoprecipitation suggests that Ssc1-3 is defective in stable binding of Mge1. We investigated this interaction further in our assay systems, using puri®ed components. As shown in Figure 4A, Mge1 was not retained when GST- Ssc1-3 protein was immobilized on the glutathione af®nity matrix, indicating that indeed Ssc1-3 could not stably bind Mge1. We then tested whether Ssc1-3 could functionally interact with Mge1, even though it did not stably bind to it. A Ssc1-3/ATP complex was isolated and tested in the single turnover assay. In contrast to the inhibition of the hydrolysis of [a-32P]ATP observed with wild-type Ssc1, addition of Mge1 had no effect on the hydrolysis of [a-32P]ATP by Ssc1-3 (Figure 4B). Also, addition of Mge1 to the Ssc1-3/ATP complex did not cause release of bound nucleotides from Ssc1-3 (Figure 4C). This inability of Mge1 to affect hydrolysis and release of prebound nucleotides indicates that Ssc1-3 neither stably binds nor functionally interacts with Mge1.

The role of a conserved loop in interaction of Ssc1 with Mge1 A previous study has identi®ed a conserved loop on the surface of DnaK which is important for interaction with GrpE (Buchberger et al., 1994a). The alteration in Ssc1-3, G79S, lies very close to the analogous loop, E56 to R62, on the modeled Ssc1 tertiary structure (the alpha-carbon atom of Figure 4. Ssc1-3 is defective in interaction with Mge1. A, G79 is approximately 10 AÊ away from the alpha- GST-Ssc1-3 fusion protein was immobilized onto gluta- carbon atom of E59). This close proximity raises thione agarose beads and puri®ed Mge1 was mixed the possibility that this corresponding loop of Ssc1 with the beads. The beads were extensively washed and eluted with 1 mM ATP as described in Materials and is important for interaction with Mge1. To test this Methods. T, Total amount of Mge1 loaded onto the hypothesis, we constructed several mutants in this beads; FT, ¯ow through fraction; SW, salt wash with region including ones analogous to those which 1 M NaCl in buffer D; B1, beads before elution with had been tested in E. coli. 56-61, corresponding adenine nucleotides; E, Mge1 eluted with adenine to 28-33 in DnaK, removesÁ the conserved loop, Á nucleotides; B2, beads after elution with adenine and G60D, corresponding to G32D in DnaK, nucleotides. B, The Ssc1-3/ATP complex ( 2 mM) was  changes a highly conserved residue within the incubated at 30C with or without 20 mM Mge1. Ali- loop (Buchberger et al., 1994a). Additional mu- quots were withdrawn at the indicated time points and tations were also constructed which result in the fraction of ATP converted to ADP determined as alterations in E59, a residue conserved between described in Materials and Methods. C, The Ssc1-3/ATP complex ( 3 mM) was incubated with or without Mge1 DnaK and Ssc1 as well as several other bacte-  (30 mM) at 30C to achieve 50 to 60% conversion of ATP rial and mitochondrial Hsp70s (Figure 5A, and to ADP before loading onto a G-50 column. Aliquots Boorstein et al., 1994). were collected and counted. Percentage of total counts The phenotype of these mutants was determined in each fraction was plotted. by testing the ability of a TRP1-based plasmid containing the mutant SSC1 to rescue growth of the strain JD100 on media containing 5-FOA. The chromosomal copy of SSC1 is disrupted in the et al., 1991b; Palleros et al., 1993) and Hsc70 (Ha & strain JD100, and the viability of the strain is McKay, 1995) maintained by a URA3-based plasmid containing a functional SSC1 gene. 5-FOA is converted to a Ssc1-3 is defective in interaction with Mge1 toxic metabolite by cells expressing the URA3 gene, thus only cells which have lost the URA3- Previously it was reported that, although Mge1 based plasmid are viable on medium containing 5- could be co-immunoprecipitated with wild-type FOA. Under these conditions, the only copy of Ssc1 from isolated mitochondria, it could not be SSC1 left in the cell is the mutant SSC1 on the co-immunoprecipitated with Ssc1-3, a tempera- TRP1-based plasmid. The growth phenotype of ture-sensitive mutant of Ssc1 (Voos et al., 1994). the resulting strain thus re¯ects the phenotype of JMB MS 1679 [15/1/97]

546 Mge1, a Nucleotide Release Factor for Ssc1

Figure 6. Phenotypes of SSC1 mutants. A, Strain JD100 was transformed with pRS314, pRS314-SSC1, or mutant SSC1 genes carried on pRS314. Transformants were streaked on plates containing 5-FOA and incubated at 30C for three days. B, The same transformants were grown in media lacking tryptophan, uracil and leucine at 30C overnight. Cells were collected and boiled in SDS-PAGE sample buffer before loading onto a SDS- Figure 5. A model of the conserved loop structure in 10% PAGE gel. An immunoblot using anti-Ssc1 anti- Ssc1. A, Sequence alignment of the loop region from body is shown. 1, Vector only; 2, wild-type SSC1;3, ssc1-3;4, 56-61;5,G60D;6,E59K;7,E59A;8,E59D. several Hsp70s. B, Tertiary structures of the loop region Á of Hsc70 and Ssc1. Data for the Hsc70 structure were obtained from the Brookhaven Data Bank. Data for the proposed Ssc1 structure were obtained from Gene Crunch, a Yeast Analysis on a Silicon Graphics affected the stable binding of Mge1 little if at all. Supercomputer (http://genecrunch.sgi.com). The struc- E59A and E59D, which were functional in vivo, tures were generated using RasMol, a program to dis- interacted with Mge1 indistinguishably from wild- play protein tertiary structures on Macintosh. type Ssc1. G60D and E59K, which did not provide Ssc1 function in vivo, were able to bind Mge1. A

the mutant SSC1. Three of the mutants, 56-61, G60D and E59K, had a null phenotype,Á as cells expressing the mutant gene were unable to form colonies at 30C and 37C (Figure 6A, and data not shown). Two of the mutants, E59A and E59D, permitted wild-type growth rates. To assure that the null phenotype observed was not due to instability of the mutant Ssc1, the level of mutant protein in cells was assessed. In the strain JD100, the functional SSC1 protein encoded by the URA3-based plasmid lacks the C-terminal 21 amino acids. Since the Ssc1 antibody is raised against a peptide corresponding to the sequence of the last 14 amino acids of Ssc1, an immunoblot using this antibody does not detect the truncated yet functional Ssc1, but rather only the mutant SSC1 protein. As shown in Figure 6B, all mutant SSC1 proteins were stable, and steady state levels Figure 7. Effects of loop mutations of Ssc1 on the stable of mutant proteins were indistinguishable from binding of Mge1. GST-Ssc1 mutant fusion proteins were wild-type Ssc1. immobilized onto glutathione agarose beads and puri- To analyze the interaction between the mutant ®ed Mge1 was mixed with the beads. The beads were SSC1 proteins and Mge1, GST-Ssc1 mutant pro- extensively washed and eluted with 1 mM ATP as teins were immobilized and tested in the stable described in Materials and Methods. T, Total amount of Mge1 loaded onto the beads; FT, ¯ow through fraction; binding assay described above. As shown in SW, salt wash with 1 M NaCl in buffer D; B , beads Figure 7, 56-61, which lacks the proposed loop 1 Á before elution with adenine nucleotides; E, Mge1 eluted structure, failed to stably bind Mge1, as expected. with adenine nucleotides; B2, beads after elution with Interestingly, the point mutants of the loop adenine nucleotides. JMB MS 1679 [15/1/97]

Mge1, a Nucleotide Release Factor for Ssc1 547

Figure 8. Effects of loop mutations of Ssc1 on its functional interaction with Mge1. Each mutant Ssc1/ATP complex ( 2 mM) was incubated at  30C with or without 20 mM Mge1. Aliquots were withdrawn at the indicated time points and the frac- tion of ATP converted to ADP determined as described in Ma- terials and Methods.

fraction of the Mge1 bound to G60D eluted with by G60D and E59K (Figure 8B and C), which the 1 M salt wash, suggesting that the ionic inter- bound Mge1 in the stable binding assay. This fail- actions between the two proteins are altered by ure to inhibit ATP hydrolysis suggests that, the mutation. When the salt concentration in the although these two mutant SSC1 proteins bind wash was lowered to 150 mM, which is more Mge1, this binding does not affect the release of physiological, all of the Mge1 remained bound to ATP from the mutant SSC1 proteins. the GST-Ssc1 mutant protein (data not shown), suggesting that the G60D is able to bind Mge1 Discussion in vivo. The portion of Mge1 that remained bound to G60D after the 1 M salt wash was eluted with Based on its sequence similarity to GrpE and its 1 mM ATP, indicating that the interaction is still interaction with Ssc1 in extracts of mitochondria, sensitive to nucleotides. Mge1 has been proposed to function as a To test the functional interaction between the nucleotide release factor for Ssc1 (Laloraya et al., mutant proteins and Mge1, we puri®ed the 1995; Westerman et al., 1995; Nakai et al., 1994). mutant proteins and tested the effect of Mge1 in The experiments reported here demonstrate that single turnover assays, as an indirect measure of Mge1 is indeed a nucleotide release factor. In a the ability of Mge1 to stimulate nucleotide release. concentration dependent manner the apparent rate Mge1 did not affect the hydrolysis of [a-32P]ATP of hydrolysis of [a-32P]ATP bound by Ssc1 is by 56-61 (Figure 8A), as expected since the two inhibited by Mge1, due to the stimulation of ATP proteinsÁ did not form a stable complex in the release by Mge1. This inhibition is nearly complete stable binding assay. However, Mge1 did affect in the presence of excess unlabeled nucleotide the hydrolysis of [a-32P]ATP bound by E59A because the released radiolabeled ATP must com- (Figure 8D) and E59D (data not shown), indicating pete for rebinding with the excess unlabeled that these changes do not interfere with the nucleotide. Analysis of the radiolabeled nucleotides Ssc1:Mge1 interaction. Interestingly, Mge1 had lit- remaining bound to and released from Ssc1 tle effect if at all on the hydrolysis of [a-32P]ATP demonstrates that the release of both ADP and ATP is stimulated by Mge1. Both Ssc1 and Mge1 are required for the proper translocation of cytosolic precursor proteins into mitochondria, as conditional mutants of either cause a delay or block of import of precursor pro- teins. How might Ssc1 and Mge1 function together Figure 9. Proposed interactions among Ssc1, Mge1 and in this process? It has been shown that the binding ATP. The proposed blocking site for the G60D and of Ssc1 to precursor proteins in transit across the E59K mutations are also shown. Ssc1 and Ssc1**, the mitochondrial membranes is crucial for the import proposed different conformations of the SSC1 protein; process (Kang et al., 1990; Gambill et al., 1993; (Ssc1*/Mge1/ATP), transient intermediate. Ungermann et al., 1994). Subsequent release of JMB MS 1679 [15/1/97]

548 Mge1, a Nucleotide Release Factor for Ssc1

imported proteins from Ssc1 allows the proper addition of Mge1 had no effect on the hydrolysis folding of these proteins, and allows the freed of ATP bound by G60D. Two explanations for the Ssc1 to function again at the import site. It has failure of Mge1 to stimulate ATP release from been proposed that the ATP-bound form of G60D are: (1) Mge1 fails to bind G60D when it is Hsp70, which binds peptides much faster, but also in the ATP-bound form; (2) Mge1 binds G60D/ releases them faster than the ADP-bound form, is ATP, but this binding does not stimulate the the Hsp70 form which actively binds polypeptides release of bound ATP. Since even wild-type Ssc1 (Greene et al., 1995; McCarty et al., 1995). Sub- does not bind Mge1 stably in the presence of ATP, sequent ATP hydrolysis converts Hsp70 to an we can not test whether G60D binds Mge1 when ADP-bound form, which has a higher af®nity for it is in the ATP-bound form. However, we were peptides, thus stabilizing the interaction. The ef®- able to test whether G60D binds Mge1 in the cient release of peptides from Hsp70 requires the ADP-bound form. While other Hsp70s have been release of bound ADP and subsequent binding of shown to retain ADP during puri®cation (Gao ATP. ADP release is stimulated by nucleotide et al., 1994; Wei & Hendershot, 1995), we were not release factors such as Mge1. Because normally the certain whether the immobilized GST-Ssc1 protein ATP concentration is higher than the ADP concen- used in the stable binding assay contained bound tration in mitochondria, Ssc1 is more likely to bind ADP. However, both wild-type Ssc1 and G60D ATP than ADP when bound nucleotide is released bound Mge1 stably in the presence of 10 mM ADP upon Mge1 binding. Therefore by facilitating (data not shown), a concentration insuf®cient to nucleotide release, Mge1 helps to more ef®ciently affect Mge1 release from Ssc1, but at least tenfold cycle Ssc1 between the ADP-bound form and the above the Kd for ADP for several Hsp70s analyzed ATP-bound form, thus facilitating cycles of poly- (Gao et al., 1993; Ha & McKay, 1994; Palleros et al., peptide binding and release. 1993; Schmid et al., 1985; Wang & Lee, 1993). Similar to the interaction of GrpE and DnaK, Mge1 must bind both the ATP-bound form and Mge1 binds Ssc1 stably. This interaction is not the ADP-bound form of wild-type Ssc1 since it disrupted by 1 M NaCl, but is sensitive to the releases both ATP and ADP from Ssc1. This addition of ATP. The Ssc1/Mge1 complex is dis- stimulation of release of both ATP and ADP by rupted at an ATP concentration as low as 10 mM. Mge1 suggests that the Mge1-binding interface of This is in excellent agreement with results Ssc1 is similar in both the ATP-bound and the obtained in isolated mitochondria, where it has ADP-bound form. Based on the ability of the been shown that Mge1 is released from Ssc1 at ADP-bound form of G60D to bind Mge1, it is 10 mM ATP but not 1 mM ATP (Bolliger et al., unlikely that the ATP-bound form of G60D is 1994). ADP is much less ef®cient in disrupting this incompetent in Mge1 binding. Therefore we pro- complex in our binding assay; 10 mM ADP was pose that the G60D mutant is defective in Mge1- required for substantial release of Mge1. Since induced nucleotide release rather than binding of Mge1 stimulates the release of both ATP and ADP Mge1. from Ssc1, and only ATP ef®ciently disrupts the Similar results were obtained for a mutant caus- Ssc1/Mge1 complex, in essence Mge1 shifts the ing an alteration at an adjacent amino acid, equilibrium to favor an ATP-bound state of Ssc1. E59K. Interestingly, E59A and E59D, two other Thus Mge1 can be thought of as a nucleotide changes at residue E59, did not affect the physi- exchange factor, since the net effect is conversion cal or functional interaction between Ssc1 and of Ssc1 from an ADP-bound state to an ATP- Mge1 in our assays. This residue is conserved bound state. between DnaK and Ssc1 (Figure 5A, see also Based on analogy to the bovine Hsc70 structure, a Boorstein et al., 1994), and has been proposed to small loop stabilized by hydrogen bonds pro- be involved in the DnaK:GrpE interaction trudes from the surface of the ATPase domain of (Buchberger et al., 1994a). However, the fact that both DnaK and Ssc1 (Figure 5B, see also Buchber- E59 can be changed to an alanine residue with- ger et al., 1994a). Deletion of the loop eliminates out affecting the Ssc1:Mge1 interaction argues the ability of Mge1 to interact with Ssc1, as well that the negative charge of E59 is not critical for as GrpE to interact with DnaK. In the case of this interaction. DnaK756, a single amino acid change within the It is intriguing that Mge1 failed to release ATP loop, G32D, prevents a stable interaction with from the two mutant proteins, G60D and E59K, GrpE. However, the analogous alteration in Ssc1, yet ATP could cause the release of Mge1. The fol- G60D, allowed stable binding of Mge1. This lowing mechanism may explain this apparent dis- apparent discrepancy between DnaK and Ssc1 crepancy. Considering the equation shown in could be due to experimental differences. The Figure 9, the release of ATP by Mge1 and the DnaK:GrpE interaction was assessed by non-dena- release of Mge1 by ATP are opposite directions of turing PAGE (Buchberger et al., 1994a), whereas the same reaction, presumably via a common tran- the Ssc1:Mge1 interaction was assessed by the sient tertiary complex of Ssc1/Mge1/ATP. We ability of Mge1 to stably associate with the im- propose that as far as the interactions among Ssc1, mobilized Ssc1. Mge1 and ATP are concerned, Ssc1 can exist in at Whereas Mge1 bound G60D, it failed to stimulate least two conformations, one with ATP bound that nucleotide release. In single turnover experiments, has a low af®nity for Mge1 (Ssc1 in Figure 9), the JMB MS 1679 [15/1/97]

Mge1, a Nucleotide Release Factor for Ssc1 549

other binds Mge1 tightly but has a low af®nity for Materials and Methods ATP (Ssc1** in Figure 9). Binding of ATP to Ssc1 will induce a conformational change in Ssc1 which Bacterial and yeast strains results in Mge1 release; binding of Mge1 will also PK101: F-, KanR, dnaKJ. Chromosomal copy of dnaK change the conformation of Ssc1, causing ATP and part of dnaJ areÁ deleted (Kang & Craig, 1990). This release. In the case of DnaK, ATP-induced confor- strain was used for expression of GST-Mge1 fusion mational changes have been well documented protein. BJ3497: pep4::HIS3 ura3-52 his 200. This strain is defec- (Banecki & Zylicz, 1996; Liberek et al., 1991b; Á Palleros et al., 1992); ¯uorescence measurements tive in Proteinase A (Jones, 1991), and was used for ex- suggest that GrpE also induces a conformational pression of GST-Ssc1 fusion proteins. JD100: lys2 ura3-52 trp1 leu2-3,112 ssc1-1(LEU2). The change in DnaK (Reid & Fink, 1996). chromosomal copy ofÁ the SSC1 gene is disrupted with According to this model, the relative concen- LEU2; Ssc1 function is provided by the truncated Ssc1 trations of ATP and Mge1 available will determine encoded on the plasmid pJD1. This truncated SSC1 pro- the favorable conformation of Ssc1 in our in vitro tein lacks the last 21 amino acids and is functional, but assays. In single turnover experiments, a large is not detected by a Ssc1 antibody which is raised excess of Mge1 was present, thus ATP was against a peptide corresponding to the last 14 amino acids of Ssc1. This strain was used to check phenotypes released from Ssc1. On the other hand, in stable as well as SSC1 protein levels of various SSC1 mutants. binding experiments, when ATP elution was per- formed, ATP was present in excess, thus Mge1 Plasmids was released from Ssc1. The two mutations on the loop, G60D and E59K, may only signi®cantly pGEXKT-MGE1: a BamHI site was generated using PCR affect one direction of the equilibrium, namely, the at nucleotides 127 to 132 (1 being the A in ATG) of release of ATP by Mge1, but not the other direc- MGE1 using the primer 50-CCCATGGGATCCGAT- tion, the release of Mge1 by ATP. Thus the bind- GAAGCCAAAAGTGAAGAATCC-30. The PCR-gener- ated fragment was either sequenced or replaced with a ing of ATP to these two mutant SSC1 proteins wild-type fragment to ensure that no PCR-induced mu- causes the conformational change necessary for tation was present. The putative mature MGE1 protein releasing Mge1, but the binding of Mge1 to the was fused to GST by cloning the PCR fragment into mutant SSC1 proteins does not cause the confor- pGEX-KT (Hakes & Dixon, 1992) as a BamHI-XhoI mational change in the mutant SSC1 proteins fragment. required for nucleotide release. pRD56CS-SSC1: a BamHI site was generated using PCR at nucleotides 70 to 75 of SSC1 (1 being the A in ATG) These defects point to a more sophisticated role of using the primer 50-ACACGTTTGGGATCCACCAA-30. the E56-R62 loop in the Ssc1:Mge1 interaction. We The PCR-generated fragment was either sequenced or propose that this loop, as well as its surrounding replaced with a wild-type fragment to ensure that no areas on the tertiary structure, are important for PCR-induced mutation was present. The mature SSC1 forming the binding interface for Mge1. Ssc1-3 protein was fused to GST by cloning the PCR fragment may signi®cantly affect this binding interface, thus into pRD56CS (pRD56 (Park et al., 1993) with ClaI and SalI sites ®lled in with Klenow) as a BamHI±EcoRI blocking the physical and functional interactions fragment. The encoded mature SSC1 protein has a gly- between the two proteins. The deletion of the loop cine residue instead of a glutamine residue at the N may have a similar effect in disrupting the terminus. Ssc1:Mge1 interaction. However, while playing a pJD1: a XhoI site was generated using PCR at role in Mge1-Ssc1 association, this loop is crucial nucleotides 1900 to 1905 of SSC1 (1 being the A in for the ability of Mge1 to induce a conformational ATG) using the primer 50-AATTATACAAGCTC- GAGTCTAACAA-30. The XhoI site was then ®lled in change of Ssc1, which leads to the release of with Klenow to generate a stop codon downstream. The bound nucleotide. One possibility is that upon encoded SSC1 protein has the last 21 amino acids Mge1 binding, the conformation of the loop replaced with LDRV (Kang, 1991). This truncated SSC1 changes, which transmits conformational changes was cloned into pRS316 as a XbaI±EcoRI fragment. to the nucleotide binding site, causing nucleotide pRS314-SSC1: wild-type SSC1 was cloned into pRS314 release. G60D and E59K may alter the confor- as a PstI±BamHI fragment. mation of the loop in a way that compromises the transmission of these changes, thus affecting Protein expression and purification nucleotide release without signi®cantly affecting Yeast strain BJ3497 harboring the expression plasmid the binding of Mge1 to Ssc1. This is the ®rst dem- pRD56CS-SSC1 was grown for two days at 30Cin onstration of the separation of Mge1 binding and 50 ml of media lacking uracil with galactose as the car- nucleotide release induced by binding of Mge1. bon source. This culture was then inoculated into one Since no analogous mutants have been found in liter of YPGal media and grown overnight at 30C. Cells DnaK, further analysis of these two mutants were harvested and resuspended in 12 ml of buffer A (16 mM Na2HPO4, 4 mM NaH2PO4, 150 mM NaCl) con- should shed more light on the mechanism of taining 1% (v/v) Triton X-100. Cells were disrupted nucleotide release from Hsp70s induced by GrpE/ with a French Pressure Cell (SLM-Aminco, Urbana, IL), Mge1. and spun at 20,000 g for 15 minutes. The soluble extract JMB MS 1679 [15/1/97]

550 Mge1, a Nucleotide Release Factor for Ssc1

was incubated with 12.5 ml of glutathione agarose fraction was then counted in a liquid scintillation coun- beads (prepared as in Lew et al., 1991) for one hour at ter to determine the amount of radioactivity associated 4C. The beads were washed extensively with buffer A with each fraction. The amount of ATP and ADP in containing 1% Triton X-100, buffer A containing 1 M the bound versus free peak was calculated from relative NaCl, buffer A, buffer B (50 mM Tris-HCl (pH 7.5), peak areas and relative amount of ATP and ADP in 50 mM NaCl, 2.5 mM CaCl2), and then 50 units of each peak fraction. At least three independent exper- thrombin (Sigma T-3010) were added. After three iments were performed with similar results; representa- hours of cleavage at 4C, the cleavage product was tive results from one experiment are shown in each collected and the beads were washed twice to col- case. lect more cleavage product trapped in the beads. The pooled cleavage product was concentrated in a Stable binding of Mge1 to Ssc1 Centriprep-10 (Amicon, Danvers, MA), adjusted to 10% (v/v) glycerol, aliquoted and stored frozen at GST-Ssc1 fusion protein was immobilized on glutathione 70C. The protein preparation was greater than agarose af®nity beads as described above. The beads 90% pure as judged from Coomassie blue staining. were washed extensively with buffer A containing 1% E. coli strain PK101 harboring the expression plasmid Triton X-100, buffer A containing 1 M NaCl, buffer A, pGEXKT-MGE1 was grown to mid log phase at 30C and buffer D (25 mM Hepes-KOH, pH 7.4, 50 mM KCl, and induced with 0.1 mM IPTG for three hours. Cells 10% glycerol, 1 mM EDTA). An equal volume of 0.1 mM were harvested and processed essentially the same as Mge1 was added to the beads and incubated at 4C for above, except the thrombin cleavage was allowed to 30 minutes. The beads were then washed with buffer D, proceed for 48 to 72 hours at 4C. The protein prep- buffer D containing 1 M NaCl, buffer D, and eluted aration was greater than 95% pure as judged from Coo- with buffer D containing 10 mM MgCl2 and the indi- massie blue staining. cated amount of ATP, ADP or various ATP analogs Mutant SSC1 proteins were puri®ed the same way as (ATP, Sigma A2383; ADP, Sigma A5136; AMPPNP, the wild-type Ssc1. Sigma A2647; ATPgS, Sigma A1388). Samples were col- lected at various stages and separated by SDS- 12% PAGE, blotted and probed with antibodies against Complex formation and single turnover experiments Mge1. ECL westerns (Amersham, Arlington Heights, IL) were performed according to the manufacturer's Ssc1 (50 mg) was incubated with [a-32P]ATP (10 mCi, suggestions. Dupont NEG-003H, 3000 Ci/mmol) in buffer C (50 mM Tris-HCl (pH 7.5), 100 mM KCl, 10 mM MgCl ,2mM 2 Site-directed mutagenesis of SSC1 DTT) containing 25 mM ATP at 30C for 15 minutes. The reaction mixture was chilled on ice and immediately Most mutants were generated using a standard two-step loaded onto a 2 ml G-50 (super®ne) column, pre-equili- PCR procedure (Cormack, 1994). The PCR fragment was brated with buffer C, at 4C. 60 ml fractions were col- sequenced to ensure the desired mutation was present lected. Fractions were monitored for radioactivity with a and no other PCR induced changes were present. For Geiger counter, and the ®rst peak of radioactivity corre- the loop deletion mutant, a linker to allow incorporation sponding to the Ssc1/ATP complex was pooled, of a glycine residue and an alanine residue was added, adjusted to 10% glycerol, aliquoted and stored frozen at as in the case of the corresponding DnaK mutant 70 C.  (Buchberger et al., 1994a). Primers used to generate the For single turnover experiments, 10 ml Ssc1/ATP com- mutants are as follows: 56-61, 50-CAAAAATTATTGC- plex was quickly thawed and added to 10 ml buffer C Á TGGTAGAACTACTCCTTCTGTAG-30 and 50-TTCTAC- containing various factors, and incubated at 30C. At CAGCAATAATTTTTGGAACTTTACCT-30; G60D, 50-A- the indicated time points, 3 ml of the reaction mixture AACGCCGAAGATTCCAGAACT-30 and 50-AGTTCTG- was withdrawn and mixed with 1 ml of stop solution GAATCTTCGGCGTTT-30; E59K, 50-TGAAAACGCCAA- containing 4 M formic acid, 2 M LiCl and 36 mM ATP. GGGTTCCAGAA-30 and 50-GTTCTGGAACCCTTGGC- This mixture was then spotted onto a PEI-cellulose TLC GTTTTC-30. Mutants E59A and E59D were generated plate (Sigma Z12,288-2) and developed in 1 M formic using a standard M13 mutagenesis procedure (Kunkel, acid and 0.5 M LiCl. The TLC plate was dried and 1985). The degenerate primer 50-AGGAGTAGTWSYG- exposed to a phosphorimager screen and data quanti- GAACCWKCGGCGTTTTC-30, where W is a A or T; S is ®ed on a phosphorimager system (Molecular Dynamics, a G or C; Y is a C or T; and K is a G or T, was used. Sunnyvale, CA). The data were plotted using Cricket Mutants were sequenced to verify the presence of the Graph. All single turnover experiments shown were per- desired mutation and the absence of other mutations. formed at least three times with similar results; repre- The mutant gene was then subcloned into pRS314-SSC1 sentative results from one experiment are shown in each for testing the phenotype and protein expression level, case. and into pRD56CS-SSC1 for testing Mge1 binding. The In the experiment where the Ssc1/ATP complex was mutant proteins were puri®ed as described above for refractionated on G-50, the Ssc1/ATP complex was use in single turnover assays. quickly thawed and incubated with or without Mge1 at 30C for short periods of time to achieve 50 to 60% conversion of [a-32P]ATP to [a-32P]ADP. The reaction mix was then chilled on ice and immediately loaded onto a 2 ml G-50 (super®ne) column pre-equilibrated Acknowledgements with buffer C at 4C. 120 ml fractions were collected. Aliquots from peak fractions were immediately mixed We are grateful to Drs Thomas Ziegelhoffer and Paul J. with stop solution and later developed on PEI-cellulose Bertics for helpful discussions and critical reading of the TLC plates to determine the relative amount of ATP manuscript, and Drs B. Diane Gambill and Hay-Oak and ADP in each fraction. An amount (50 ml) of each Park for gifts of strains and plasmids. This work was JMB MS 1679 [15/1/97]

Mge1, a Nucleotide Release Factor for Ssc1 551

supported by the NIH grant 5RO1 GM27870 to E. A. C. of actin and the ATPase fragment of a 70kDa heat and NIH predoctoral training grant NRSA 5T32 shock cognate protein. Proc. Natl Acad. Sci. USA, 88, GM07215 to J. E. D. 5041±5045. Flynn, G. C., Chappell, T. G. & Rothman, J. E. (1989). Peptide binding and release by proteins implicated References as catalysts of protein assembly. Science, 245, 385± 390. Ahsen, O. v., Voos, W., Henninger, H. & Pfanner, N. Freeman, B., Myers, M., Schumacher, R. & Morimoto, R. (1995). The mitochondrial protein import machin- (1995). Identi®cation of a regulatory motif in Hsp70 ery. Role of ATP in dissociation of the Hsp70/ that affects ATPase activity, substrate binding and Mim44 complex. J. Biol. Chem. 270, 29848±29853. interaction with HDJ-1. EMBO J. 14, 2281±2292. Banecki, B. & Zylicz, M. (1996). Real time kinetics of the Gambill, B. D., Voos, W., Kang, P. J., Miao, B., Langer, DnaK/DnaJ/GrpE molecular chaperone machine T., Craig, E. A. & Pfanner, N. (1993). A dual role action. J. Biol. Chem. 271, 6137±6143. for mitochondrial 70 in mem- Blond-Elguindi, S., Cwirla, S., Dower, W., Lipshutz, R., brane translocation of preproteins. J. Cell Biol. 123, Sprang, S., Sambrook, J. & Gething, M. J. (1993). 109±117. Af®nity panning of a library of peptides displayed Gao, B., Emoto, Y., Greene, L. & Eisenberg, E. (1993). on reveals the binding speci®city of Nucleotide binding properties of bovine brain BiP. Cell, 75, 717±728. uncoating ATPase. J. Biol. Chem. 268, 8507±8513. Bolliger, L., Deloche, O., Glick, B., Georgopoulos, C., Gao, B., Greene, L. & Eisenberg, E. (1994). Characteriz- Jeno, P., Kronidou, N., Horst, M., Morishima, N. & ation of nucleotide-free uncoating ATPase and its Schatz, G. (1994). A mitochondrial homolog of bac- binding to ATP, ADP, and ATP analogues. Biochem- terial GrpE interacts with mitochondrial hsp70 and istry, 33, 2048±2054. is essential for viability. EMBO J. 13, 1998±2006. Georgopoulos, C., Liberek, K., Zylicz, M. & Ang, D. Boorstein, W. R., Ziegelhoffer, T. & Craig, E. A. (1994). (1994) . Properties of the heat shock proteins of Molecular evolution of the HSP70 multigene family. Escherichia coli and the autoregulation of the heat J. Mol. Evol. 38, 1±17. shock response. In The Biology of Heat Shock Proteins Bork, P., Sander, C. & Valencia, A. (1992). An ATPase and Molecular Chaperones (Morimoto, R. I., Tissieres, domain common to procaryotic cell cycle proteins, A. & Georgopoulos, C., eds.), pp. 209±250, Cold sugar kinases, actin and hsp70 heat shock proteins. Spring Harbor Laboratory Press, Plainview, NY. Proc. Natl Acad. Sci. USA, 89, 7290±7294. Gragerov, A., Zeng, L., Zhao, X., Burkholder, W. & Buchberger, A., Schroder, H., Buttner, M., Valencia, A. & Gottesman, M. E. (1994). Speci®city of DnaK-pep- Bukau, B. (1994a). A conserved loop in the ATPase tide binding. J. Mol. Biol. 235, 848±854. domain of the DnaK chaperone is essential for Greene, L., Zinner, R., Na®cy, S. & Eisenberg, E. (1995). stable binding of GrpE. Nature: Struct. Biol. 1,95± Effect of nucleotide on the binding of peptides to 101. 70-kDa heat shock proteins. J. Biol. Chem. 270, Buchberger, A., Valencia, A., McMacken, R., Sander, 2967±2973. C. & Bukau, B. (1994b). The chaperone function of Ha, J.-H. & McKay, D. (1994). ATPase kinetics of recom- DnaK requires the coupling of ATPase activity with binant bovine 70kDa heat shock cognate protein substrate binding through residue E171. EMBO J. and its amino-terminal ATPase domain. Biochemis- 13, 1687±1695. try, 33, 14625±14635. Buchberger, A., Theyssen, H., Schroder, H., McCarty, Ha, J. H. & McKay, D. B. (1995). Kinetics of nucleotide- J. S., Virgallita, G., Milkereit, P., Reinstein, J. & induced changes in the tryptophan ¯uorescence of Bukau, B. (1995). Nucleotide-induced conformation- the molecular chaperone Hsc70 and its subfrag- al changes in the ATPase and substrate binding ments suggest the ATP-induced conformational domains of the DnaK chaperone provide evidence change follows initial ATP binding. Biochemistry, 34, for interdomain communication. J. Biol. Chem. 270, 16903±16910. 11635±44. Chappell, T. G., Konforti, B. B., Schmid, S. L. & Hakes, D. J. & Dixon, J. E. (1992). New vectors for high Rothman, J. E. (1987). The ATPase core of a cla- level expression of recombinant proteins in . thrin uncoating protein. J. Biol. Chem. 262, 746±751. Anal. Biochem. 202, 293±298. Cormack, B. (1994). Introduction of a point mutation by Hartl, F. U. (1996). Molecular chaperones in cellular pro- sequential PCR steps. In Current Protocols in Molecu- tein folding. Nature, 381, 571±580. lar Biology (Ausubel, F. M., Brent, R., Kingston, R. E., Horst, M., Oppliger, W., Feifel, B., Schatz, G. & Glick, B. Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, (1996). The mitochondrial import motor: dis- K., eds.), vol. 1, pp. 8.5.7±8.5.9, John Wiley & Sons, sociation of mitochondrial hsp70 from its membrane New York. anchor requires ATP binding rather than ATP Craig, E. A., Gambill, B. D. & Nelson, R. J. (1993). Heat hydrolysis. Protein Sci. 5, 759±767. shock proteins: molecular chaperones of protein Ikeda, E., Yoshida, S., Mitsuzawa, H., Uno, I. & Toh-e, biogenesis. Microbiol. Rev. 57, 402±414. A. (1994). YGE1 is a yeast homologue of Escherichia Cyr, D. M., Langer, T. & Douglas, M. G. (1994). DnaJ- coli grpE and is required for maintenance of mito- like proteins: molecular chaperones and speci®c reg- chondrial function. FEBS Letters, 339, 265±268. ulators of Hsp70. Trends Biochem. Sci. 19, 176±181. Jones, E. W. (1991). Tackling the problem in Flaherty, K. M., DeLuca-Flaherty, C. & McKay, D. B. Saccharomyces cerevisiae. Methods Enzymol. 194, 428± (1990). Three dimensional structure of the ATPase 453. fragment of a 70k heat-shock cognate protein. Jordan, R. & McMacken, R. (1995). Modulation of the Nature, 346, 623±628. ATPase activity of the molecular chaperone DnaK Flaherty, K. M., McKay, D. B., Kabash, W. & Holmes, K. by peptides and the DnaJ and GrpE heat shock (1991). Similarity of the three-dimensional structure proteins. J. Biol. Chem, 270, 4563±4569. JMB MS 1679 [15/1/97]

552 Mge1, a Nucleotide Release Factor for Ssc1

Kang, P. J. (1991). Functional studies of 70 kDa heat teins in transit through mitochondrial contact sites shock proteins in E. coli and yeast. Doctoral interact with hsp70 in the matrix. FEBS Letters, 277, dissertation, University of Wisconsin-Madison, WI. 281±284. Kang, P. J. & Craig, E. A. (1990). Identi®cation and Palleros, D. R., Reid, K. L., McCarty, J. S., Walker, characterization of a new Escherichia coli gene that is G. C. & Fink, A. L. (1992). DnaK, hsp73, and their a dosage-dependent suppressor of a dnaK deletion molten globules. J. Biol. Chem. 267, 5279±5285. mutation. J. Bacteriol. 172, 2055±2064. Palleros, D. R., Reid, K. L., Shi, L., Welch, W. & Fink, Kang, P. J., Ostermann, J., Shilling, J., Neupert, W., A. L. (1993). ATP-induced protein-hsp70 complex Craig, E. A. & Pfanner, N. (1990). Hsp70 in the dissociation requires K‡ but not ATP hydrolysis. mitochondrial matrix is required for translocation Nature, 365, 664±666. and folding of precursor proteins. Nature, 348, 137± Park, H. O., Chant, J. & Herskowitz, I. (1993). BUD2 143. encodes a GTPase-activating protein for Bud1/Rsr1 Kunkel, T. A. (1985). Rapid and ef®cient site-speci®c necessary for proper bud-site selection in yeast. mutagenesis without phenotypic selection. Proc. Nature, 365, 269±274. Natl Acad. Sci. USA, 82, 488±492. Reid, K. L. & Fink, A. L. (1996). Physical interactions Laloraya, S., Gambill, B. D. & Craig, E. A. (1994). A role between members of the DnaK chaperone machin- for a eukaryotic GrpE-related protein, Mge1p, in ery: characterization of the DnaK.GrpE complex. protein translocation. Proc. Natl Acad. Sci. USA, 91, Cell Stress and Chaperones, 1, 127±137. 6481±6485. Scherer, P., Krieg, U., Hwang, S., Vestweber, D. & Laloraya, S., Dekker, P., Voos, W., Craig, E. & Pfanner, Schatz, G. (1990). A precursor protein partially N. (1995). Mitochondrial GrpE modulates the func- translocated into yeast mitochondria is bound to a tion of matrix Hsp70 in translocation and matu- 70 kDa mitochondrial stress protein. EMBO J. 9, ration of proteins. Mol. Cell. Biol. 15, 7098±7105. 4315±4322. Lew, A. M., Beck, D. J. & Thomas, L. M. (1990). Recom- Schmid, S. L., Braell, W. A. & Rothman, J. E. (1985). binant fusion proteins of protein A and protein G ATP catalyzes the sequestration of clathrin during with glutathione S-transferase as reporter proteins. enzymatic uncoating. J. Biol. Chem. 260, 10057± J. Immunol. Methods, 136, 211±219. 10062. Liberek, K., Marszalek, J., Ang, D. & Georgopoulos, C. Ungermann, C., Neupert, W. & Cyr, D. M. (1994). The (1991a). Escherichia coli DnaJ and GrpE heat shock role of Hsp70 in conferring unidirectionality on pro- proteins jointly stimulate ATPase activity of DnaK. tein translocation into mitochondria. Science, 266, Proc. Natl Acad. Sci. USA, 88, 2874±2878. 1250±1253. Liberek, K., Skowyra, D., Zylicz, M., Johnson, C. & Voos, W., Gambill, B. D., Laloraya, S., Ang, D., Craig, Georgopoulos, C. (1991b). The Escherichia coli DnaK E. A. & Pfanner, N. (1994). Mitochondrial GrpE is chaperone, the 70-kDa heat shock protein eukary- present in a complex with hsp70 and preproteins in otic equivalent, changes conformation upon ATP transit across membranes. Mol. Cell. Biol. 14, 6627± hydrolysis, thus triggering its dissociation from a 6634.. bound target protein. J. Biol. Chem. 266, 14491± Wang, C. & Lee, M. R. (1993). High-level expression of 14496. soluble rat hsc70 in Escherichia coli: puri®cation and Manning-Krieg, U. C., Scherer, P. & Schatz, G. (1991). characterization of the cloned enzyme. Biochem. J. Sequential action of mitochondrial chaperones in 294, 69±77. protein import into the matrix. EMBO J. 10, 3273± Wang, T.-F., Chang, J.-h. & Wang, C. (1993). Identi®- 3280. cation of the peptide binding domain of hsc70. McCarty, J., Buchberger, A., Reinstein, J. & Bukau, B. J. Biol. Chem. 268, 26049±26051. (1995). The role of ATP in the functional cycle of Wei, J. & Hendershot, L. (1995). Characterization of the the DnaK chaperone system. J. Mol. Biol. 249, 126± nucleotide binding properties and ATPase activity 137. of recombinant hamster BiP puri®ed from bacteria. Morimoto, R. I., Tissieres, A. & Georgopoulos, C. (1994). J. Biol. Chem. 270, 26670±26676. The Biology of Heat Shock Proteins and Molecular Westerman, B., Prip-Buus, C., Neupert, W. & Schwarz, Chaperones . Cold Spring Harbor Monograph Series E. (1995). The role of the GrpE homologue, Mge1p, Cold Spring Harbor Laboratory Press, Cold Spring in mediating protein import and protein folding in Harbor, New York. mitochondria. EMBO J. 13, 1998±2006. Nakai, M., Kato, Y., Ikeda, E., Toh-e, A. & Endo, T. Zhu, X., Zhao, X., Burkholder, W., Gragerov, A., Ogata, (1994). YGE1p, a eukaryotic GrpE homologue, is C., Gottesman, M. & Hendrickson, W. (1996). Struc- localized in the mitochondrial matrix and interacts tural analysis of substrate binding by the molecular with mitochondrial hsp70. Biochem. Biophys. Res. chaperone DnaK. Science, 272, 1606±1614. Commun. 200, 435±441. Zylicz, M., Ang, D. & Georgopoulos, C. (1987). The Ostermann, J., Voos, W., Kang, P. J., Craig, E. A., GrpE protein of Escherichia coli. J. Biol. Chem. 262, Neupert, W. & Pfanner, N. (1990). Precursor pro- 17437±17442.

Edited by M. Gottesman

(Received 10 July 1996; received in revised form 10 October 1996; accepted 5 November 1996)